An object tracking method based on Mean Shift algorithm with HSV color space and texture features

Mean Shift is a powerful and versatile non-parametric iterative algorithm that can be used for lot of purposes like finding modes, clustering etc. It has been widely used in target tracking field because of some advantages like fewer iteration times and better real-time performance for many years. However, due to only single-color histogram representation of target feature has been used in traditional Mean Shift algorithm, it cannot track very well in some cases, especially under very complicated conditions. There are mainly two problems that can cause traditional Mean Shift algorithm to be unstable. The first problem is when the background color and target color are similar, the tracking performance is significantly insufficient, the second is the partial occlusion problem. In this paper, we have proposed a solution to solve these two issues, which contains three improvements. For the first problem, we transformed original color features in traditional Mean Shift algorithm into HSV color space, At the same time, we will also consider texture features and integrate into algorithm to improve tracking performance. and, we applied four neighborhood searching method to solve partial occlusion problem. We tested our algorithms on a variety of standard datasets and a video data collected from real-world environment. The result of experiments show that our proposed algorithm has higher accuracy than traditional Mean Shift algorithm and the background weighted Mean Shift algorithm in test case of complex conditions. Besides, our proposed algorithm also has a good operating efficiency then traditional one.